This chapter will be largely a summary of the special evidence bearing upon the different hypotheses that have been offered for the explanation of such phenomena as we have obtained. I will take up the principal hypotheses, one at a time, with a summarized regrouping of the evidence that bears particularly on the evaluation of the given hypothesis. The detailed conditions will mostly be omitted, since they have been given in Part II, along with the presentation of the results. These main hypotheses have been referred to occasionally in Part II, in passing. Doubtless, most readers will already be convinced of their inapplicability by this time. In such case, this chapter may be omitted and the thread taken up at Chapter 10. Logically, the first alternative suggestion that is evoked to explain unusual results such as these high scores in card-guessing is that it "just happened". That is, that no special principle of causation is responsible; rather, that a number of unimportant circumstances contributed the peculiar results. This general absence of a special causal principle we call "accident" or "chance", and we mean merely that no recognized general causal principle is responsible. This we can call the Chance Hypothesis. According to the Chance Hypothesis, we would be as likely to go below chance average, if we ran 90,000 more trials, as we would be to go above. All the positive deviation we have accumulated has just been one grand, persistent accident, stretching through three years of varied conditions and over a wide range of subjects. It has actually been suggested to me by a colleague urging this hypothesis that I will some day find my results swinging as far in a negative deviation as they have already in the positive. What, then, can we say to this? First, there is the mathematical evaluative principle of probability, by which we may be sure of the odds against an event occurring by chance alone. Since this principle is used throughout Part II and is explained at the end of Chapter 2, it is unnecessary to go into its explanation here. It is only necessary to repeat that general practise among statisticians relies upon a deviation of 4 times the probable error (of the mean expectation, n.p.) for the minimum limit of significance to reveal the operation of a general factor that is something more than mere accident or chance. This is arbitrary (but one might require 3 or 5 as his minimum value for X [i.e., X=d/p.e.] if he feels it is warranted.). With X=4, the odds If one wishes to test out the mathematical value for probability, it is easy to do. I have, myself, conducted 4,000 "chance" trials, by first making and recording the calls, and later shuffling the cards and checking them against the recorded calls. This eliminated clairvoyance; I did not even try to think of any particular pack of cards. Chance average of 4,000 trials is 800. I got 801. For the last 1,000 of these I took the scores made by Pearce, who was then averaging around 10 hits per 25 trials; I took the same packs of cards he had been using (about 12 in number), cut them once each time and checked them against the record. Chance expectation was, of course, 200 for 1,000 trials and Pearce had almost doubled this, getting 386 correct. My "chance" control series, however, gave only 204 or a deviation of less than .5 of the probable error. Of course, the mathematical theory has been tested many times and it would be a waste of time to go further into testing it for our purpose. Also, the minimal value of 4 for significance in X seems, according to this brief testing, amply adequate and relatively conservative. Most people are more impressed by a spectacular series of successive hits than by lower but cumulative scoring. Pearce's scoring 25 straight under clairvoyant conditions, in my presence, and Zirkle's 26 straight hits in pure telepathy with my assistant, Miss Ownbey, are the best instances of these. Other subjects have approached these. Linzmayer scored 21 in 25 with clairvoyance, in my presence; Miss Ownbey herself, unwitnessed, scored 23, pure clairvoyance. Miss Turner's score of 19 in distance P.T. work stands out because of the 250 miles between her and the agent. Miss Bailey scored 19 in P.T. in the same room with the agent, as did also Cooper. The odds against getting one series of 25 straight hits by mere chance would be 525 which is nearly 300 quadrillions—just one score of 25! A small part of our 90,000 trials. Once we are certain that we are not dealing with a mere accidental deviation—that the Chance Hypothesis has been adequately ruled out Table XIX in Chapter VII. Among these seven witnesses are three psychologists, an education official and a professional magician. The very magnitude of the system of persons involved, whose names are herein published, must discourage any attempt at a charge of sheer dishonesty. Perhaps I may, for brevity's sake, refer the possible doubting reader to Dr. McDougall or Dr. Prince (both of whom have known me now for many years) concerning such a point. All the major subjects, themselves, have been witnessed to some extent and most of them almost entirely. Much of this witnessing has been done by trusted graduate assistants, young research students who are themselves going in for a psychological career and are fully responsible. The very division of this responsibility offers in itself a more complex obstacle to the "Fraud Hypothesis". Each one witnessed results of ample significance to prove E.S.P. on their own merit. Those subjects who have not been wholly witnessed throughout have in many cases shown better results when witnessed than when unwitnessed (see Table XXVIII, for one illustration; Stuart is another instance). And, finally, on the point of witnessing of subjects, there are several long and very significant series of tests by major subjects doubly witnessed, i.e., with two of us present. See, for example, Table XIX, Columns headed B and C. Note that Column C totals with a deviation over 23 times the p.e.—doubly witnessed figures. The honesty of the subjects does not matter under such conditions. But, in my judgment of them, these subjects are all splendidly In view of the simple technique used and the relatively simple computations required, it can hardly be seriously thought that the results herein reported are the consequence of errors made by the observers. Moreover, on this point, Mr. Pratt and I together witnessed several hundred of Mr. Pearce's best scores. Dr. Zener and I together witnessed 300 trials with Pearce, observing all points of the procedure together and using new cards. The deviation of the results is 9.4 times the p.e. for these 300 trials. None of the various observers have been able to point to any adequate weakness or combination of weaknesses in the procedure that could, in their judgment, explain the results. Some suggestions have been made for further improving the technique but no adequate loophole discovered. Again, the independent recording of the able assistants and myself is a check, to a certain extent, upon the competence and reliability of us all. Moreover, the early results, obtained when we were all less experienced, were among the very poorest. With improvement in technique and judgment, the results have risen in value. This does not look like incompetence. Furthermore, the natural outlines of the data, the curves, the fluctuations with physiological conditions, the effects of the presence of strange witnesses, of changes, of illness, all show a lawfulness that gets us considerably beyond this "Hypothesis of Haphazard Observation". It is true, much of this work was of a tentative, exploratory character and lacked standardization. It is also probable that some errors have been made in recording, totalling and computing the values. If so, such errors are at most of trifling consequence. The general ground has been covered too often and by too many individuals for serious error that would vitiate an important conclusion. So the Hypothesis of Incompetence will, I think, find few adherents and no justification.
There is left, then, when we safely pass the above mentioned hypotheses, that of Unconscious Sensory Perception. That is, assuming that the investigator is honest and competent enough to pass muster, and that the subject is guarded carefully enough to prevent outright dishonesty on his part (if it should exist), might there not be sensory indices such as marks on the back of the card, peculiarities on the edge of the card, unconscious whispering from agent to perceiver in P.T. work and the like? This is unconscious sensory perception, since, if the subject is honest and is conscious, he would not presumably do this, knowing he was deceiving. But the conditions for prevention of deception, conscious and unconscious, are just about the same and the two may be discussed together. Beginning with Linzmayer, there were 120 trials made with combined P.C. and P.T. conditions (i.e., I held a card and looked at it also), with screened card, subject's face turned away and a motor going to cover possible unconscious whispering. Results gave a deviation 11.2 times the p.e. With pure clairvoyance in 105 trials, excluding sensory perception by screening with 55 and by using new cards with 50, the deviation is, alone, 7 times the p.e. Turning now to Pearce, we find first that the new card data, totalling 1,675 trials, yielded a positive deviation of 291 (±11) and X=26. These averaged 9.3 in 25. The trials included the 1st, 2nd and 3rd runs with the new cards. The first runs averaged as high as the 3rd. During these runs Pearce did not look at the cards before he called them, as a rule. He would glance at the pack now and then, however, absent-mindedly. He called each card before lifting it off. There had been no chance for him to learn any indication marks on the backs of the cards, since he never saw them before the period of experimentation, and during that time he did not see the face and back of a card consecutively. The 600 trials made with a screen between the cards and the subject, Pearce, are also highly significant, both the B.T. and the 300 with the possible telepathic factor included. The significance indices (X) are, respectively, 8.3 and 11.9. The D.T. data, obtained with nothing but the edges of the cards visible, except the top card, afford good exclusion of sensory perception. Most of the time Pearce did not even look at the pack after starting the calling. But, even when he did, apparently uniform edges of the cards offered no clue to their identity. Often they were new cards, and never were they so old that differences in battered edges could be detected and associated with the face. Moreover, there was no chance to learn the connection. The calls were made rapidly (25 in 1 to 2 minutes) and the Combining these data and in view of their significance, what sensory perception is possible? No one present at the tests or absent, has been able to suggest a possibility. Did this young minister sneak into my laboratory and "thumb nail" the edges of the cards for the D.T. work? Aside from the fact of the other data (screen and new cards, etc.), and the not unimportant facts that my laboratory was kept locked and that Pearce did not usually look at the pack during D.T., there is the simple obstacle that I, too, can see such marks and have looked for them frequently during the thousands of hours I have spent in this work. I have never discovered marks that might have been purposely made, except once, on the backs of some of one old pack of cards, and these were not consistent. They may well have been nervously and absent-mindedly thumb-nailed by an idle observer. In one lot of cards, also, Stuart showed me that the rectangles were on slightly broader cards than were the other figures. Thereafter we had the cards cut better and more evenly. On the telepathic side there are the distance studies, which get well beyond the range of sensory perception of the unconscious order. However, in all our later and better P.T. work done in one room we kept an electric fan going. The noise of this motor is enough to "drown" any supposed unconscious whispering. Also, in the excellent P.T. work of Miss Turner, Miss Bailey, Cooper and Zirkle the percipients did not look at the agent. There was, then, no sensory contact between the two, except for the uniform tapping of a telegraph key, giving the signal for each call to be made. In the distance work, the distance was simply added to these other conditions, and a tile wall or two interposed along with distance. Sensory perception is simply eliminated from the case. And, as a climax, there is the long distance telepathy of Miss Turner, over 250 miles away from the agent, Miss Ownbey, and the already very successful distance P.C. work of Pearce from one building to another on the campus. For the details of these various P.T. scores and totals under the various conditions, consult Table XXXVII, in Chapter 8. The results are all splendidly rich in significance. In fact, the anti-chance index, X, for the distance P.T. data alone on Misses Bailey and Turner, and for Cooper and Zirkle is 46, an enormous value. There is, I think, no point to elaborating further on the Sensory Perception Hypothesis. There can hardly be any just doubt of its successful What are the possibilities for rational inference? That is, how successfully can the subject determine by reasoning what card is on top of a given pack or at any other point in it? It is almost obvious that in a shuffled and cut pack of cards, containing 5 each of 5 suits, no logic known can determine the distribution. And, in our experience, the effort to use logic only interferes with success in E.S.P. When the cards are called and checked in 5's, as was often done, there might seem to be a chance for the subject to remember to some advantage in the last 5 calls the number of various circles, rectangles, etc., already called. But our good subjects did not depend on inference and did not attempt to use it. If this had been a factor, one would expect the last 5 calls in the B.T.-by-5's to be relatively higher than the B.T.-by-25's (i.e., where all were called before checking up). This is not the case. In fact, the opposite is true. While Pearce comes in heavily in all P.C. work on the last 5 calls, he gets a much higher relative score on the last 5 of the B.T. 25. He does also in the D.T. (which is also run straight through by 25's). So we may see that this seemingly possible slight weakness is not a measurably actual one. Even theoretically, however, there could be an inference made only in case all of one suit had been checked out in the first 20 cards of the pack and this was noted by the subject. Then the chance of getting each call correct would be raised from 1/5 to 1/4. This could be an error, then only to that slight extent. Actually, the subject who kept adding and checking up rationally thus would not have the state of abstraction necessary to get above chance average. We may, then, discard the Hypothesis of Rational Inference from the P.C. end—as a matter of fact, only a small portion of the 75,000 or more P.C. trials were run as B.T. 5, and then, frequently, after each five the cards would be returned to the pack and reshuffled for the next five calls. But, turning to the P.T. phase, we have a more difficult problem. In order to get free from possible clairvoyance, we avoid the use of cards or objective record of any kind, except that made after the call is given. The agent chooses, without objective aid, what symbols to visualize in imagination. We decided that the best way to avoid trouble on this score was to have the agent choose not just one figure at a time, but a series of 5. The order of the five would be determined upon and would be followed out, concentrating imaginal attention upon each one as its turn came. After these five were used, another five would be mentally selected, usually some variation of the first five. The same five symbols as are used on the Zener cards are used throughout the P.T. work. The degree of repetition and variation is measured roughly by the range one gets from the average pack, and the percipient is told to expect that. Thus a system is followed by the agent, yet a system that varies continually, and thus avoids stereotyped order that might be inferred or fallen into by accident. A study of the agents’ records shows no detectable order, except the orderly avoidance of order. The percipient was not told by the agent of his success until the end of the run of 25. Thus he had no basis for inference as he went. Here, again, the rational attitude is destructive, as in P.C. work. And here, too, we may dismiss the Hypothesis of Rational Inference. At its best, no one could suppose it capable of explaining the brilliant long runs, the 25 straight successes, under both P.C. and P.T. conditions. The hypotheses discussed here are the main ones that conflict or compete with the E.S.P. hypothesis. They have been, one by one, eliminated and, since they have no special strength in combination, we may conclude that E.S.P. stands without a serious opposing hypothesis. The evidence for the elimination of these opposing hypotheses was such that it eliminated each one completely, not leaving a partial support for any one of them. On this ground, too, then, we may omit discussion of combinations of the opposing hypotheses. For those, then, who can accept proof before explanation is arrived at (i.e., for the scientifically mature) E.S.P. is a natural fact and principle, puzzling as its explanation may be. We can turn now to the facts that tend to throw light upon its nature and functioning. 112:1 And to this we can now add the excellent distance-P.C. results of Pearce, mentioned at the close of Chapters 3 and 7.CHAPTER 9
Elimination of Negative Hypotheses
(a) THE HYPOTHESIS OF CHANCE
(b) FRAUD HYPOTHESIS
(C) HYPOTHESIS OF INCOMPETENCE
Footnotes